Deep Learning in Radiology: Does One Size Fit All?

Bradley J Erickson, Panagiotis Korfiatis, Timothy Kline, Zeynettin Akkus, Kenneth Philbrick, Alexander D. Weston

Research output: Contribution to journalArticle

21 Citations (Scopus)

Abstract

Deep learning (DL) is a popular method that is used to perform many important tasks in radiology and medical imaging. Some forms of DL are able to accurately segment organs (essentially, trace the boundaries, enabling volume measurements or calculation of other properties). Other DL networks are able to predict important properties from regions of an image-for instance, whether something is malignant, molecular markers for tissue in a region, even prognostic markers. DL is easier to train than traditional machine learning methods, but requires more data and much more care in analyzing results. It will automatically find the features of importance, but understanding what those features are can be a challenge. This article describes the basic concepts of DL systems and some of the traps that exist in building DL systems and how to identify those traps.

Original languageEnglish (US)
JournalJournal of the American College of Radiology
DOIs
StateAccepted/In press - Jan 1 2018

Fingerprint

Radiology
Learning
Diagnostic Imaging

Keywords

  • Computer-aided diagnosis
  • Deep learning
  • Machine learning

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Cite this

Deep Learning in Radiology : Does One Size Fit All? / Erickson, Bradley J; Korfiatis, Panagiotis; Kline, Timothy; Akkus, Zeynettin; Philbrick, Kenneth; Weston, Alexander D.

In: Journal of the American College of Radiology, 01.01.2018.

Research output: Contribution to journalArticle

@article{e1a13e8c8fa24111b5161bb9670522c6,
title = "Deep Learning in Radiology: Does One Size Fit All?",
abstract = "Deep learning (DL) is a popular method that is used to perform many important tasks in radiology and medical imaging. Some forms of DL are able to accurately segment organs (essentially, trace the boundaries, enabling volume measurements or calculation of other properties). Other DL networks are able to predict important properties from regions of an image-for instance, whether something is malignant, molecular markers for tissue in a region, even prognostic markers. DL is easier to train than traditional machine learning methods, but requires more data and much more care in analyzing results. It will automatically find the features of importance, but understanding what those features are can be a challenge. This article describes the basic concepts of DL systems and some of the traps that exist in building DL systems and how to identify those traps.",
keywords = "Computer-aided diagnosis, Deep learning, Machine learning",
author = "Erickson, {Bradley J} and Panagiotis Korfiatis and Timothy Kline and Zeynettin Akkus and Kenneth Philbrick and Weston, {Alexander D.}",
year = "2018",
month = "1",
day = "1",
doi = "10.1016/j.jacr.2017.12.027",
language = "English (US)",
journal = "JACR Journal of the American College of Radiology",
issn = "1558-349X",
publisher = "Elsevier BV",

}

TY - JOUR

T1 - Deep Learning in Radiology

T2 - Does One Size Fit All?

AU - Erickson, Bradley J

AU - Korfiatis, Panagiotis

AU - Kline, Timothy

AU - Akkus, Zeynettin

AU - Philbrick, Kenneth

AU - Weston, Alexander D.

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Deep learning (DL) is a popular method that is used to perform many important tasks in radiology and medical imaging. Some forms of DL are able to accurately segment organs (essentially, trace the boundaries, enabling volume measurements or calculation of other properties). Other DL networks are able to predict important properties from regions of an image-for instance, whether something is malignant, molecular markers for tissue in a region, even prognostic markers. DL is easier to train than traditional machine learning methods, but requires more data and much more care in analyzing results. It will automatically find the features of importance, but understanding what those features are can be a challenge. This article describes the basic concepts of DL systems and some of the traps that exist in building DL systems and how to identify those traps.

AB - Deep learning (DL) is a popular method that is used to perform many important tasks in radiology and medical imaging. Some forms of DL are able to accurately segment organs (essentially, trace the boundaries, enabling volume measurements or calculation of other properties). Other DL networks are able to predict important properties from regions of an image-for instance, whether something is malignant, molecular markers for tissue in a region, even prognostic markers. DL is easier to train than traditional machine learning methods, but requires more data and much more care in analyzing results. It will automatically find the features of importance, but understanding what those features are can be a challenge. This article describes the basic concepts of DL systems and some of the traps that exist in building DL systems and how to identify those traps.

KW - Computer-aided diagnosis

KW - Deep learning

KW - Machine learning

UR - http://www.scopus.com/inward/record.url?scp=85041133227&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85041133227&partnerID=8YFLogxK

U2 - 10.1016/j.jacr.2017.12.027

DO - 10.1016/j.jacr.2017.12.027

M3 - Article

C2 - 29396120

AN - SCOPUS:85041133227

JO - JACR Journal of the American College of Radiology

JF - JACR Journal of the American College of Radiology

SN - 1558-349X

ER -